net/ipv4/ping.c

   1 /*
   2  * INET         An implementation of the TCP/IP protocol suite for the LINUX
   3  *              operating system.  INET is implemented using the  BSD Socket
   4  *              interface as the means of communication with the user level.
   5  *
   6  *              "Ping" sockets
   7  *
   8  *              This program is free software; you can redistribute it and/or
   9  *              modify it under the terms of the GNU General Public License
  10  *              as published by the Free Software Foundation; either version
  11  *              2 of the License, or (at your option) any later version.
  12  *
  13  * Based on ipv4/udp.c code.
  14  *
  15  * Authors:     Vasiliy Kulikov / Openwall (for Linux 2.6),
  16  *              Pavel Kankovsky (for Linux 2.4.32)
  17  *
  18  * Pavel gave all rights to bugs to Vasiliy,
  19  * none of the bugs are Pavel's now.
  20  *
  21  */
  22
  23 #include <asm/system.h>
  24 #include <linux/uaccess.h>
  25 #include <linux/types.h>
  26 #include <linux/fcntl.h>
  27 #include <linux/socket.h>
  28 #include <linux/sockios.h>
  29 #include <linux/in.h>
  30 #include <linux/errno.h>
  31 #include <linux/timer.h>
  32 #include <linux/mm.h>
  33 #include <linux/inet.h>
  34 #include <linux/netdevice.h>
  35 #include <net/snmp.h>
  36 #include <net/ip.h>
  37 #include <net/ipv6.h>
  38 #include <net/icmp.h>
  39 #include <net/protocol.h>
  40 #include <linux/skbuff.h>
  41 #include <linux/proc_fs.h>
  42 #include <net/sock.h>
  43 #include <net/ping.h>
  44 #include <net/icmp.h>
  45 #include <net/udp.h>
  46 #include <net/route.h>
  47 #include <net/inet_common.h>
  48 #include <net/checksum.h>
  49
  50
  51 static struct ping_table ping_table;
  52
  53 static u16 ping_port_rover;
  54
  55 static inline int ping_hashfn(struct net *net, unsigned num, unsigned mask)
  56 {
  57         int res = (num + net_hash_mix(net)) & mask;
  58         pr_debug("hash(%d) = %d\n", num, res);
  59         return res;
  60 }
  61
  62 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
  63                                              struct net *net, unsigned num)
  64 {
  65         return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
  66 }
  67
  68 static int ping_v4_get_port(struct sock *sk, unsigned short ident)
  69 {
  70         struct hlist_nulls_node *node;
  71         struct hlist_nulls_head *hlist;
  72         struct inet_sock *isk, *isk2;
  73         struct sock *sk2 = NULL;
  74
  75         isk = inet_sk(sk);
  76         write_lock_bh(&ping_table.lock);
  77         if (ident == 0) {
  78                 u32 i;
  79                 u16 result = ping_port_rover + 1;
  80
  81                 for (i = 0; i < (1L << 16); i++, result++) {
  82                         if (!result)
  83                                 result++; /* avoid zero */
  84                         hlist = ping_hashslot(&ping_table, sock_net(sk),
  85                                             result);
  86                         ping_portaddr_for_each_entry(sk2, node, hlist) {
  87                                 isk2 = inet_sk(sk2);
  88
  89                                 if (isk2->inet_num == result)
  90                                         goto next_port;
  91                         }
  92
  93                         /* found */
  94                         ping_port_rover = ident = result;
  95                         break;
  96 next_port:
  97                         ;
  98                 }
  99                 if (i >= (1L << 16))
 100                         goto fail;
 101         } else {
 102                 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
 103                 ping_portaddr_for_each_entry(sk2, node, hlist) {
 104                         isk2 = inet_sk(sk2);
 105
 106                         if ((isk2->inet_num == ident) &&
 107                             (sk2 != sk) &&
 108                             (!sk2->sk_reuse || !sk->sk_reuse))
 109                                 goto fail;
 110                 }
 111         }
 112
 113         pr_debug("found port/ident = %d\n", ident);
 114         isk->inet_num = ident;
 115         if (sk_unhashed(sk)) {
 116                 pr_debug("was not hashed\n");
 117                 sock_hold(sk);
 118                 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
 119                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 120         }
 121         write_unlock_bh(&ping_table.lock);
 122         return 0;
 123
 124 fail:
 125         write_unlock_bh(&ping_table.lock);
 126         return 1;
 127 }
 128
 129 static void ping_v4_hash(struct sock *sk)
 130 {
 131         pr_debug("ping_v4_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
 132         BUG(); /* "Please do not press this button again." */
 133 }
 134
 135 static void ping_v4_unhash(struct sock *sk)
 136 {
 137         struct inet_sock *isk = inet_sk(sk);
 138         pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
 139         if (sk_hashed(sk)) {
 140                 struct hlist_nulls_head *hslot;
 141
 142                 hslot = ping_hashslot(&ping_table, sock_net(sk), isk->inet_num);
 143                 write_lock_bh(&ping_table.lock);
 144                 hlist_nulls_del(&sk->sk_nulls_node);
 145                 sock_put(sk);
 146                 isk->inet_num = isk->inet_sport = 0;
 147                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
 148                 write_unlock_bh(&ping_table.lock);
 149         }
 150 }
 151
 152 static struct sock *ping_v4_lookup(struct net *net, u32 saddr, u32 daddr,
 153                                    u16 ident, int dif)
 154 {
 155         struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
 156         struct sock *sk = NULL;
 157         struct inet_sock *isk;
 158         struct hlist_nulls_node *hnode;
 159
 160         pr_debug("try to find: num = %d, daddr = %ld, dif = %d\n",
 161                          (int)ident, (unsigned long)daddr, dif);
 162         read_lock_bh(&ping_table.lock);
 163
 164         ping_portaddr_for_each_entry(sk, hnode, hslot) {
 165                 isk = inet_sk(sk);
 166
 167                 pr_debug("found: %p: num = %d, daddr = %ld, dif = %d\n", sk,
 168                          (int)isk->inet_num, (unsigned long)isk->inet_rcv_saddr,
 169                          sk->sk_bound_dev_if);
 170
 171                 pr_debug("iterate\n");
 172                 if (isk->inet_num != ident)
 173                         continue;
 174                 if (isk->inet_rcv_saddr && isk->inet_rcv_saddr != daddr)
 175                         continue;
 176                 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
 177                         continue;
 178
 179                 sock_hold(sk);
 180                 goto exit;
 181         }
 182
 183         sk = NULL;
 184 exit:
 185         read_unlock_bh(&ping_table.lock);
 186
 187         return sk;
 188 }
 189
 190 static void inet_get_ping_group_range_net(struct net *net, gid_t *low,
 191                                           gid_t *high)
 192 {
 193         gid_t *data = net->ipv4.sysctl_ping_group_range;
 194         unsigned seq;
 195         do {
 196                 seq = read_seqbegin(&sysctl_local_ports.lock);
 197
 198                 *low = data[0];
 199                 *high = data[1];
 200         } while (read_seqretry(&sysctl_local_ports.lock, seq));
 201 }
 202
 203
 204 static int ping_init_sock(struct sock *sk)
 205 {
 206         struct net *net = sock_net(sk);
 207         gid_t group = current_egid();
 208         gid_t range[2];
 209         struct group_info *group_info = get_current_groups();
 210         int i, j, count = group_info->ngroups;
 211
 212         inet_get_ping_group_range_net(net, range, range+1);
 213         if (range[0] <= group && group <= range[1])
 214                 return 0;
 215
 216         for (i = 0; i < group_info->nblocks; i++) {
 217                 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
 218
 219                 for (j = 0; j < cp_count; j++) {
 220                         group = group_info->blocks[i][j];
 221                         if (range[0] <= group && group <= range[1])
 222                                 return 0;
 223                 }
 224
 225                 count -= cp_count;
 226         }
 227
 228         return -EACCES;
 229 }
 230
 231 static void ping_close(struct sock *sk, long timeout)
 232 {
 233         pr_debug("ping_close(sk=%p,sk->num=%u)\n",
 234                 inet_sk(sk), inet_sk(sk)->inet_num);
 235         pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
 236
 237         sk_common_release(sk);
 238 }
 239
 240 /*
 241  * We need our own bind because there are no privileged id's == local ports.
 242  * Moreover, we don't allow binding to multi- and broadcast addresses.
 243  */
 244
 245 static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 246 {
 247         struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
 248         struct inet_sock *isk = inet_sk(sk);
 249         unsigned short snum;
 250         int chk_addr_ret;
 251         int err;
 252
 253         if (addr_len < sizeof(struct sockaddr_in))
 254                 return -EINVAL;
 255
 256         pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
 257                 sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
 258
 259         chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
 260         if (addr->sin_addr.s_addr == INADDR_ANY)
 261                 chk_addr_ret = RTN_LOCAL;
 262
 263         if ((sysctl_ip_nonlocal_bind == 0 &&
 264             isk->freebind == 0 && isk->transparent == 0 &&
 265              chk_addr_ret != RTN_LOCAL) ||
 266             chk_addr_ret == RTN_MULTICAST ||
 267             chk_addr_ret == RTN_BROADCAST)
 268                 return -EADDRNOTAVAIL;
 269
 270         lock_sock(sk);
 271
 272         err = -EINVAL;
 273         if (isk->inet_num != 0)
 274                 goto out;
 275
 276         err = -EADDRINUSE;
 277         isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
 278         snum = ntohs(addr->sin_port);
 279         if (ping_v4_get_port(sk, snum) != 0) {
 280                 isk->inet_saddr = isk->inet_rcv_saddr = 0;
 281                 goto out;
 282         }
 283
 284         pr_debug("after bind(): num = %d, daddr = %ld, dif = %d\n",
 285                 (int)isk->inet_num,
 286                 (unsigned long) isk->inet_rcv_saddr,
 287                 (int)sk->sk_bound_dev_if);
 288
 289         err = 0;
 290         if (isk->inet_rcv_saddr)
 291                 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
 292         if (snum)
 293                 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
 294         isk->inet_sport = htons(isk->inet_num);
 295         isk->inet_daddr = 0;
 296         isk->inet_dport = 0;
 297         sk_dst_reset(sk);
 298 out:
 299         release_sock(sk);
 300         pr_debug("ping_v4_bind -> %d\n", err);
 301         return err;
 302 }
 303
 304 /*
 305  * Is this a supported type of ICMP message?
 306  */
 307
 308 static inline int ping_supported(int type, int code)
 309 {
 310         if (type == ICMP_ECHO && code == 0)
 311                 return 1;
 312         return 0;
 313 }
 314
 315 /*
 316  * This routine is called by the ICMP module when it gets some
 317  * sort of error condition.
 318  */
 319
 320 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 321
 322 void ping_err(struct sk_buff *skb, u32 info)
 323 {
 324         struct iphdr *iph = (struct iphdr *)skb->data;
 325         struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
 326         struct inet_sock *inet_sock;
 327         int type = icmph->type;
 328         int code = icmph->code;
 329         struct net *net = dev_net(skb->dev);
 330         struct sock *sk;
 331         int harderr;
 332         int err;
 333
 334         /* We assume the packet has already been checked by icmp_unreach */
 335
 336         if (!ping_supported(icmph->type, icmph->code))
 337                 return;
 338
 339         pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
 340                 code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 341
 342         sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
 343                             ntohs(icmph->un.echo.id), skb->dev->ifindex);
 344         if (sk == NULL) {
 345                 ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
 346                 pr_debug("no socket, dropping\n");
 347                 return; /* No socket for error */
 348         }
 349         pr_debug("err on socket %p\n", sk);
 350
 351         err = 0;
 352         harderr = 0;
 353         inet_sock = inet_sk(sk);
 354
 355         switch (type) {
 356         default:
 357         case ICMP_TIME_EXCEEDED:
 358                 err = EHOSTUNREACH;
 359                 break;
 360         case ICMP_SOURCE_QUENCH:
 361                 /* This is not a real error but ping wants to see it.
 362                  * Report it with some fake errno. */
 363                 err = EREMOTEIO;
 364                 break;
 365         case ICMP_PARAMETERPROB:
 366                 err = EPROTO;
 367                 harderr = 1;
 368                 break;
 369         case ICMP_DEST_UNREACH:
 370                 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
 371                         if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
 372                                 err = EMSGSIZE;
 373                                 harderr = 1;
 374                                 break;
 375                         }
 376                         goto out;
 377                 }
 378                 err = EHOSTUNREACH;
 379                 if (code <= NR_ICMP_UNREACH) {
 380                         harderr = icmp_err_convert[code].fatal;
 381                         err = icmp_err_convert[code].errno;
 382                 }
 383                 break;
 384         case ICMP_REDIRECT:
 385                 /* See ICMP_SOURCE_QUENCH */
 386                 err = EREMOTEIO;
 387                 break;
 388         }
 389
 390         /*
 391          *      RFC1122: OK.  Passes ICMP errors back to application, as per
 392          *      4.1.3.3.
 393          */
 394         if (!inet_sock->recverr) {
 395                 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
 396                         goto out;
 397         } else {
 398                 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
 399                          info, (u8 *)icmph);
 400         }
 401         sk->sk_err = err;
 402         sk->sk_error_report(sk);
 403 out:
 404         sock_put(sk);
 405 }
 406
 407 /*
 408  *      Copy and checksum an ICMP Echo packet from user space into a buffer.
 409  */
 410
 411 struct pingfakehdr {
 412         struct icmphdr icmph;
 413         struct iovec *iov;
 414         u32 wcheck;
 415 };
 416
 417 static int ping_getfrag(void *from, char * to,
 418                         int offset, int fraglen, int odd, struct sk_buff *skb)
 419 {
 420         struct pingfakehdr *pfh = (struct pingfakehdr *)from;
 421
 422         if (offset == 0) {
 423                 if (fraglen < sizeof(struct icmphdr))
 424                         BUG();
 425                 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
 426                             pfh->iov, 0, fraglen - sizeof(struct icmphdr),
 427                             &pfh->wcheck))
 428                         return -EFAULT;
 429
 430                 return 0;
 431         }
 432         if (offset < sizeof(struct icmphdr))
 433                 BUG();
 434         if (csum_partial_copy_fromiovecend
 435                         (to, pfh->iov, offset - sizeof(struct icmphdr),
 436                          fraglen, &pfh->wcheck))
 437                 return -EFAULT;
 438         return 0;
 439 }
 440
 441 static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
 442                                     struct flowi4 *fl4)
 443 {
 444         struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
 445
 446         pfh->wcheck = csum_partial((char *)&pfh->icmph,
 447                 sizeof(struct icmphdr), pfh->wcheck);
 448         pfh->icmph.checksum = csum_fold(pfh->wcheck);
 449         memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
 450         skb->ip_summed = CHECKSUM_NONE;
 451         return ip_push_pending_frames(sk, fl4);
 452 }
 453
 454 static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 455                         size_t len)
 456 {
 457         struct net *net = sock_net(sk);
 458         struct flowi4 fl4;
 459         struct inet_sock *inet = inet_sk(sk);
 460         struct ipcm_cookie ipc;
 461         struct icmphdr user_icmph;
 462         struct pingfakehdr pfh;
 463         struct rtable *rt = NULL;
 464         struct ip_options_data opt_copy;
 465         int free = 0;
 466         u32 saddr, daddr, faddr;
 467         u8  tos;
 468         int err;
 469
 470         pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
 471
 472
 473         if (len > 0xFFFF)
 474                 return -EMSGSIZE;
 475
 476         /*
 477          *      Check the flags.
 478          */
 479
 480         /* Mirror BSD error message compatibility */
 481         if (msg->msg_flags & MSG_OOB)
 482                 return -EOPNOTSUPP;
 483
 484         /*
 485          *      Fetch the ICMP header provided by the userland.
 486          *      iovec is modified!
 487          */
 488
 489         if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
 490                              sizeof(struct icmphdr)))
 491                 return -EFAULT;
 492         if (!ping_supported(user_icmph.type, user_icmph.code))
 493                 return -EINVAL;
 494
 495         /*
 496          *      Get and verify the address.
 497          */
 498
 499         if (msg->msg_name) {
 500                 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
 501                 if (msg->msg_namelen < sizeof(*usin))
 502                         return -EINVAL;
 503                 if (usin->sin_family != AF_INET)
 504                         return -EINVAL;
 505                 daddr = usin->sin_addr.s_addr;
 506                 /* no remote port */
 507         } else {
 508                 if (sk->sk_state != TCP_ESTABLISHED)
 509                         return -EDESTADDRREQ;
 510                 daddr = inet->inet_daddr;
 511                 /* no remote port */
 512         }
 513
 514         ipc.addr = inet->inet_saddr;
 515         ipc.opt = NULL;
 516         ipc.oif = sk->sk_bound_dev_if;
 517         ipc.tx_flags = 0;
 518         err = sock_tx_timestamp(sk, &ipc.tx_flags);
 519         if (err)
 520                 return err;
 521
 522         if (msg->msg_controllen) {
 523                 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
 524                 if (err)
 525                         return err;
 526                 if (ipc.opt)
 527                         free = 1;
 528         }
 529         if (!ipc.opt) {
 530                 struct ip_options_rcu *inet_opt;
 531
 532                 rcu_read_lock();
 533                 inet_opt = rcu_dereference(inet->inet_opt);
 534                 if (inet_opt) {
 535                         memcpy(&opt_copy, inet_opt,
 536                                sizeof(*inet_opt) + inet_opt->opt.optlen);
 537                         ipc.opt = &opt_copy.opt;
 538                 }
 539                 rcu_read_unlock();
 540         }
 541
 542         saddr = ipc.addr;
 543         ipc.addr = faddr = daddr;
 544
 545         if (ipc.opt && ipc.opt->opt.srr) {
 546                 if (!daddr)
 547                         return -EINVAL;
 548                 faddr = ipc.opt->opt.faddr;
 549         }
 550         tos = RT_TOS(inet->tos);
 551         if (sock_flag(sk, SOCK_LOCALROUTE) ||
 552             (msg->msg_flags & MSG_DONTROUTE) ||
 553             (ipc.opt && ipc.opt->opt.is_strictroute)) {
 554                 tos |= RTO_ONLINK;
 555         }
 556
 557         if (ipv4_is_multicast(daddr)) {
 558                 if (!ipc.oif)
 559                         ipc.oif = inet->mc_index;
 560                 if (!saddr)
 561                         saddr = inet->mc_addr;
 562         }
 563
 564         flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
 565                            RT_SCOPE_UNIVERSE, sk->sk_protocol,
 566                            inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
 567
 568         security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
 569         rt = ip_route_output_flow(net, &fl4, sk);
 570         if (IS_ERR(rt)) {
 571                 err = PTR_ERR(rt);
 572                 rt = NULL;
 573                 if (err == -ENETUNREACH)
 574                         IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
 575                 goto out;
 576         }
 577
 578         err = -EACCES;
 579         if ((rt->rt_flags & RTCF_BROADCAST) &&
 580             !sock_flag(sk, SOCK_BROADCAST))
 581                 goto out;
 582
 583         if (msg->msg_flags & MSG_CONFIRM)
 584                 goto do_confirm;
 585 back_from_confirm:
 586
 587         if (!ipc.addr)
 588                 ipc.addr = fl4.daddr;
 589
 590         lock_sock(sk);
 591
 592         pfh.icmph.type = user_icmph.type; /* already checked */
 593         pfh.icmph.code = user_icmph.code; /* ditto */
 594         pfh.icmph.checksum = 0;
 595         pfh.icmph.un.echo.id = inet->inet_sport;
 596         pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
 597         pfh.iov = msg->msg_iov;
 598         pfh.wcheck = 0;
 599
 600         err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
 601                         0, &ipc, &rt, msg->msg_flags);
 602         if (err)
 603                 ip_flush_pending_frames(sk);
 604         else
 605                 err = ping_push_pending_frames(sk, &pfh, &fl4);
 606         release_sock(sk);
 607
 608 out:
 609         ip_rt_put(rt);
 610         if (free)
 611                 kfree(ipc.opt);
 612         if (!err) {
 613                 icmp_out_count(sock_net(sk), user_icmph.type);
 614                 return len;
 615         }
 616         return err;
 617
 618 do_confirm:
 619         dst_confirm(&rt->dst);
 620         if (!(msg->msg_flags & MSG_PROBE) || len)
 621                 goto back_from_confirm;
 622         err = 0;
 623         goto out;
 624 }
 625
 626 static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 627                         size_t len, int noblock, int flags, int *addr_len)
 628 {
 629         struct inet_sock *isk = inet_sk(sk);
 630         struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
 631         struct sk_buff *skb;
 632         int copied, err;
 633
 634         pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
 635
 636         if (flags & MSG_OOB)
 637                 goto out;
 638
 639         if (addr_len)
 640                 *addr_len = sizeof(*sin);
 641
 642         if (flags & MSG_ERRQUEUE)
 643                 return ip_recv_error(sk, msg, len);
 644
 645         skb = skb_recv_datagram(sk, flags, noblock, &err);
 646         if (!skb)
 647                 goto out;
 648
 649         copied = skb->len;
 650         if (copied > len) {
 651                 msg->msg_flags |= MSG_TRUNC;
 652                 copied = len;
 653         }
 654
 655         /* Don't bother checking the checksum */
 656         err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 657         if (err)
 658                 goto done;
 659
 660         sock_recv_timestamp(msg, sk, skb);
 661
 662         /* Copy the address. */
 663         if (sin) {
 664                 sin->sin_family = AF_INET;
 665                 sin->sin_port = 0 /* skb->h.uh->source */;
 666                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
 667                 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 668         }
 669         if (isk->cmsg_flags)
 670                 ip_cmsg_recv(msg, skb);
 671         err = copied;
 672
 673 done:
 674         skb_free_datagram(sk, skb);
 675 out:
 676         pr_debug("ping_recvmsg -> %d\n", err);
 677         return err;
 678 }
 679
 680 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 681 {
 682         pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
 683                 inet_sk(sk), inet_sk(sk)->inet_num, skb);
 684         if (sock_queue_rcv_skb(sk, skb) < 0) {
 685                 ICMP_INC_STATS_BH(sock_net(sk), ICMP_MIB_INERRORS);
 686                 kfree_skb(skb);
 687                 pr_debug("ping_queue_rcv_skb -> failed\n");
 688                 return -1;
 689         }
 690         return 0;
 691 }
 692
 693
 694 /*
 695  *      All we need to do is get the socket.
 696  */
 697
 698 void ping_rcv(struct sk_buff *skb)
 699 {
 700         struct sock *sk;
 701         struct net *net = dev_net(skb->dev);
 702         struct iphdr *iph = ip_hdr(skb);
 703         struct icmphdr *icmph = icmp_hdr(skb);
 704         u32 saddr = iph->saddr;
 705         u32 daddr = iph->daddr;
 706
 707         /* We assume the packet has already been checked by icmp_rcv */
 708
 709         pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
 710                 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 711
 712         /* Push ICMP header back */
 713         skb_push(skb, skb->data - (u8 *)icmph);
 714
 715         sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
 716                             skb->dev->ifindex);
 717         if (sk != NULL) {
 718                 pr_debug("rcv on socket %p\n", sk);
 719                 ping_queue_rcv_skb(sk, skb_get(skb));
 720                 sock_put(sk);
 721                 return;
 722         }
 723         pr_debug("no socket, dropping\n");
 724
 725         /* We're called from icmp_rcv(). kfree_skb() is done there. */
 726 }
 727
 728 struct proto ping_prot = {
 729         .name =         "PING",
 730         .owner =        THIS_MODULE,
 731         .init =         ping_init_sock,
 732         .close =        ping_close,
 733         .connect =      ip4_datagram_connect,
 734         .disconnect =   udp_disconnect,
 735         .setsockopt =   ip_setsockopt,
 736         .getsockopt =   ip_getsockopt,
 737         .sendmsg =      ping_sendmsg,
 738         .recvmsg =      ping_recvmsg,
 739         .bind =         ping_bind,
 740         .backlog_rcv =  ping_queue_rcv_skb,
 741         .hash =         ping_v4_hash,
 742         .unhash =       ping_v4_unhash,
 743         .get_port =     ping_v4_get_port,
 744         .obj_size =     sizeof(struct inet_sock),
 745 };
 746 EXPORT_SYMBOL(ping_prot);
 747
 748 #ifdef CONFIG_PROC_FS
 749
 750 static struct sock *ping_get_first(struct seq_file *seq, int start)
 751 {
 752         struct sock *sk;
 753         struct ping_iter_state *state = seq->private;
 754         struct net *net = seq_file_net(seq);
 755
 756         for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
 757              ++state->bucket) {
 758                 struct hlist_nulls_node *node;
 759                 struct hlist_nulls_head *hslot;
 760
 761                 hslot = &ping_table.hash[state->bucket];
 762
 763                 if (hlist_nulls_empty(hslot))
 764                         continue;
 765
 766                 sk_nulls_for_each(sk, node, hslot) {
 767                         if (net_eq(sock_net(sk), net))
 768                                 goto found;
 769                 }
 770         }
 771         sk = NULL;
 772 found:
 773         return sk;
 774 }
 775
 776 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
 777 {
 778         struct ping_iter_state *state = seq->private;
 779         struct net *net = seq_file_net(seq);
 780
 781         do {
 782                 sk = sk_nulls_next(sk);
 783         } while (sk && (!net_eq(sock_net(sk), net)));
 784
 785         if (!sk)
 786                 return ping_get_first(seq, state->bucket + 1);
 787         return sk;
 788 }
 789
 790 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
 791 {
 792         struct sock *sk = ping_get_first(seq, 0);
 793
 794         if (sk)
 795                 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
 796                         --pos;
 797         return pos ? NULL : sk;
 798 }
 799
 800 static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
 801 {
 802         struct ping_iter_state *state = seq->private;
 803         state->bucket = 0;
 804
 805         read_lock_bh(&ping_table.lock);
 806
 807         return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
 808 }
 809
 810 static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 811 {
 812         struct sock *sk;
 813
 814         if (v == SEQ_START_TOKEN)
 815                 sk = ping_get_idx(seq, 0);
 816         else
 817                 sk = ping_get_next(seq, v);
 818
 819         ++*pos;
 820         return sk;
 821 }
 822
 823 static void ping_seq_stop(struct seq_file *seq, void *v)
 824 {
 825         read_unlock_bh(&ping_table.lock);
 826 }
 827
 828 static void ping_format_sock(struct sock *sp, struct seq_file *f,
 829                 int bucket, int *len)
 830 {
 831         struct inet_sock *inet = inet_sk(sp);
 832         __be32 dest = inet->inet_daddr;
 833         __be32 src = inet->inet_rcv_saddr;
 834         __u16 destp = ntohs(inet->inet_dport);
 835         __u16 srcp = ntohs(inet->inet_sport);
 836
 837         seq_printf(f, "%5d: %08X:%04X %08X:%04X"
 838                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
 839                 bucket, src, srcp, dest, destp, sp->sk_state,
 840                 sk_wmem_alloc_get(sp),
 841                 sk_rmem_alloc_get(sp),
 842                 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
 843                 atomic_read(&sp->sk_refcnt), sp,
 844                 atomic_read(&sp->sk_drops), len);
 845 }
 846
 847 static int ping_seq_show(struct seq_file *seq, void *v)
 848 {
 849         if (v == SEQ_START_TOKEN)
 850                 seq_printf(seq, "%-127s\n",
 851                            "  sl  local_address rem_address   st tx_queue "
 852                            "rx_queue tr tm->when retrnsmt   uid  timeout "
 853                            "inode ref pointer drops");
 854         else {
 855                 struct ping_iter_state *state = seq->private;
 856                 int len;
 857
 858                 ping_format_sock(v, seq, state->bucket, &len);
 859                 seq_printf(seq, "%*s\n", 127 - len, "");
 860         }
 861         return 0;
 862 }
 863
 864 static const struct seq_operations ping_seq_ops = {
 865         .show           = ping_seq_show,
 866         .start          = ping_seq_start,
 867         .next           = ping_seq_next,
 868         .stop           = ping_seq_stop,
 869 };
 870
 871 static int ping_seq_open(struct inode *inode, struct file *file)
 872 {
 873         return seq_open_net(inode, file, &ping_seq_ops,
 874                            sizeof(struct ping_iter_state));
 875 }
 876
 877 static const struct file_operations ping_seq_fops = {
 878         .open           = ping_seq_open,
 879         .read           = seq_read,
 880         .llseek         = seq_lseek,
 881         .release        = seq_release_net,
 882 };
 883
 884 static int ping_proc_register(struct net *net)
 885 {
 886         struct proc_dir_entry *p;
 887         int rc = 0;
 888
 889         p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
 890         if (!p)
 891                 rc = -ENOMEM;
 892         return rc;
 893 }
 894
 895 static void ping_proc_unregister(struct net *net)
 896 {
 897         proc_net_remove(net, "icmp");
 898 }
 899
 900
 901 static int __net_init ping_proc_init_net(struct net *net)
 902 {
 903         return ping_proc_register(net);
 904 }
 905
 906 static void __net_exit ping_proc_exit_net(struct net *net)
 907 {
 908         ping_proc_unregister(net);
 909 }
 910
 911 static struct pernet_operations ping_net_ops = {
 912         .init = ping_proc_init_net,
 913         .exit = ping_proc_exit_net,
 914 };
 915
 916 int __init ping_proc_init(void)
 917 {
 918         return register_pernet_subsys(&ping_net_ops);
 919 }
 920
 921 void ping_proc_exit(void)
 922 {
 923         unregister_pernet_subsys(&ping_net_ops);
 924 }
 925
 926 #endif
 927
 928 void __init ping_init(void)
 929 {
 930         int i;
 931
 932         for (i = 0; i < PING_HTABLE_SIZE; i++)
 933                 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
 934         rwlock_init(&ping_table.lock);
 935 }